Biography 🚀

Trying to make silicon do what carbon can. Download my resumé .

Interests

  • Machine Learning
  • Deep Learning
  • Computer Vision

Education

  • MS in Data Science, 2021

    New York University

  • B.Tech in Information Technology, 2019

    Vellore Institute of Technology

Experience

 
 
 
 
 

Machine Learning Intern

PepsiCo

Jun 2020 – Aug 2020 New York, NY, USA
  • Trained a Word2Vec-like model on customer product baskets to project ~10,000 unique products to a low-dimensional embedding space (using TensorFlow).
  • Evaluated learned representations using the measures of “complementarity” and “substitutability” between different products, inspired by the paper - Product2Vec.
  • Used t-SNE to visualize and qualitatively validate the embedding similarity between complementary/substitute products.
 
 
 
 
 

Data Science Intern

Noodle.ai

Jan 2019 – May 2019 Bengaluru, India
  • Developed a modular and reusable python framework for working with sequence-to-sequence models (using Keras).
  • Designed a benchmark suite for evaluating different sequence-to-sequence architectures for multivariate time-series forecasting (on internal datasets of the company).
  • Standardized and adapted the data preprocessing pipeline across various ongoing projects to handle time-series data.
 
 
 
 
 

Data Science Intern

Noodle.ai

May 2018 – Jun 2018 Bengaluru, India
  • Worked on a system to estimate the time required by an order (of steel coils) to get ready. Performed exploratory data analysis on the client’s data with the aim to understand the manufacturing process and model it.
  • Further, created an auxiliary conditional probabilistic model for a specific scenario in the manufacturing process. When integrated into the existing system, this model could improve the results by ~20%.
  • Made extensive use of Python (along with various open-source libraries including, but not limited to Pandas, NumPy, Scikit-learn and Plotly) for data pre-processing, modeling and data visualization.

Projects

Head and Neck CT image segmentation using deep learning

Segmenting organs in Head and Neck CT scans
PDF Code
Head and Neck CT image segmentation using deep learning

Generating bird’s-eye view from multi-image scene using self-supervised learning

Self-supervised learning in traffic environment
PDF Code
Generating bird’s-eye view from multi-image scene using self-supervised learning

Neural Style Transfer

Implemented neural style transfer with numerous extensions
PDF Code
Neural Style Transfer

Mydia

A simple and efficient wrapper for reading videos as NumPy tensors.
Code
Mydia

Contact